用tensorflow构建线性回归模型的示例代码

#encoding=utf-8 
__author__ = 'freedom' 
import tensorflow as tf 
import numpy as np 
 
def createData(dataNum,w,b,sigma): 
 train_x = np.arange(dataNum) 
 train_y = w*train_x+b+np.random.randn()*sigma 
 #print train_x 
 #print train_y 
 return train_x,train_y 
 
def linerRegression(train_x,train_y,epoch=100000,rate = 0.000001): 
 train_x = np.array(train_x) 
 train_y = np.array(train_y) 
 n = train_x.shape[0] 
 x = tf.placeholder("float") 
 y = tf.placeholder("float") 
 w = tf.Variable(tf.random_normal([1])) # 生成随机权重 
 b = tf.Variable(tf.random_normal([1])) 
 
 pred = tf.add(tf.mul(x,w),b) 
 loss = tf.reduce_sum(tf.pow(pred-y,2)) 
 optimizer = tf.train.GradientDescentOptimizer(rate).minimize(loss) 
 init = tf.initialize_all_variables() 
 
 sess = tf.Session() 
 sess.run(init) 
 print 'w start is ',sess.run(w) 
 print 'b start is ',sess.run(b) 
 for index in range(epoch): 
  #for tx,ty in zip(train_x,train_y): 
   #sess.run(optimizer,{x:tx,y:ty}) 
  sess.run(optimizer,{x:train_x,y:train_y}) 
  # print 'w is ',sess.run(w) 
  # print 'b is ',sess.run(b) 
  # print 'pred is ',sess.run(pred,{x:train_x}) 
  # print 'loss is ',sess.run(loss,{x:train_x,y:train_y}) 
  #print '------------------' 
 print 'loss is ',sess.run(loss,{x:train_x,y:train_y}) 
 w = sess.run(w) 
 b = sess.run(b) 
 return w,b 
 
def predictionTest(test_x,test_y,w,b): 
 W = tf.placeholder(tf.float32) 
 B = tf.placeholder(tf.float32) 
 X = tf.placeholder(tf.float32) 
 Y = tf.placeholder(tf.float32) 
 n = test_x.shape[0] 
 pred = tf.add(tf.mul(X,W),B) 
 loss = tf.reduce_mean(tf.pow(pred-Y,2)) 
 sess = tf.Session() 
 loss = sess.run(loss,{X:test_x,Y:test_y,W:w,B:b}) 
 return loss 
 
if __name__ == "__main__": 
 train_x,train_y = createData(50,2.0,7.0,1.0) 
 test_x,test_y = createData(20,2.0,7.0,1.0) 
 w,b = linerRegression(train_x,train_y) 
 print 'weights',w 
 print 'bias',b 
 loss = predictionTest(test_x,test_y,w,b) 
 print loss